Question Marker Drop in Japanese

Japanese has been known as a language that utilizes morphological question markers to indicate interrogative sentences (cf. Cheng, 1991). It has been pointed out, however, that question markers in Japanese can be optionally omitted as shown in sentences like (1) and (2) below (cf. Inoue, 1996 and Lasnik & Saito, 1992):

Although this question marker drop (QM-drop) phenomenon is commonly observed in informal speech, it is not the case that question markers can be omitted freely without any constraints. One of our goals in this paper is to describe the QM-Drop phenomenon and its major restrictions. Our second goal is to provide a possible explanation of this phenomenon within the framework of Principles and Parameters approach. Based on cross-linguistic comparison with well-studied languages, we argue that question markers may be omitted only if interrogative features of the sentence can be properly licensed. Adopting a standard assumption that the interrogative force is carried by C⁰ with the LF interpretable feature [+wh] or [+Q], we claim that there are three basic ways to license the interrogative feature in Japanese: (i) by overt realization of the [+wh] or [+Q] feature with a morphological question marker, (ii) by an instance of Spec-head agreement of [+wh] feature (i.e., "dynamic agreement" of Rizzi, 1996), and (iii) by I-to-C head-movement.

After suggesting that I-to-C head-movement may be at work in Japanese as well as languages like English and French (cf. Subject-Aux Inversion), we discuss two theoretically important consequences from our analysis of QM-drop. One is that I-to-C head-movement in Japanese may not take place in overt syntax in the embedded context. The other is that the S-structure null operator movement (cf. Watanabe, 1992 and Takahashi, 1993; 1995), if it exists at all, is not enough to license the interrogative feature in C⁰ (i,e., for the purpose of "clause typing" in the sense of Cheng (1991).